MTF
Public Types | Public Member Functions | Static Public Member Functions | Public Attributes | List of all members
PFParams Struct Reference

Public Types

enum  DynamicModel { RandomWalk, AutoRegression1 }
 
enum  UpdateType { Additive, Compositional }
 
enum  ResamplingType { None, BinaryMultinomial, LinearMultinomial, Residual }
 
enum  LikelihoodFunc { AM, Gaussian, Reciprocal }
 
enum  MeanType { None, SSM, Corners }
 

Public Member Functions

 PFParams (int _max_iters, int _n_particles, double _epsilon, DynamicModel _dyn_model, UpdateType _upd_type, LikelihoodFunc _likelihood_func, ResamplingType _resampling_type, MeanType _mean_type, bool _reset_to_mean, const vectorvd &_ssm_sigma, const vectorvd &_ssm_mean, bool _update_distr_wts, double _min_distr_wt, double _adaptive_resampling_thresh, const vectord &_pix_sigma, double _measurement_sigma, int _show_particles, bool _enable_learning, bool _jacobian_as_sigma, bool _debug_mode)
 
 PFParams (const PFParams *params=nullptr)
 
bool processDistributions (vector< VectorXd > &state_sigma, vector< VectorXd > &state_mean, VectorXi &distr_n_samples, unsigned int &n_distr, unsigned int ssm_state_size)
 parse the provided mean and sigma and apply several priors to get the final parameters for all distributions
 

Static Public Member Functions

static const char * toString (DynamicModel _dyn_model)
 
static const char * toString (UpdateType _upd_type)
 
static const char * toString (ResamplingType _resampling_type)
 
static const char * toString (LikelihoodFunc _likelihood_func)
 
static const char * toString (MeanType _likelihood_func)
 

Public Attributes

int max_iters
 maximum iterations of the PF algorithm to run for each frame
 
int n_particles
 number of particles to use
 
double epsilon
 iterations will be terminated when L2 norm of the change in tracker corners exceeds this
 
DynamicModel dynamic_model
 
UpdateType update_type
 
LikelihoodFunc likelihood_func
 
ResamplingType resampling_type
 
MeanType mean_type
 method used for computing the mean of the SSM states corresponding to the particles. More...
 
bool reset_to_mean
 reset all particles to the mean/optimal corners found in each frame
 
vectorvd ssm_sigma
 standarsd deviations of the Gaussian distributions to use for the samplers
 
vectorvd ssm_mean
 mean of the Gaussian distributions to use for the samplers
 
bool update_distr_wts
 update the proportion of samples taken from different sampler according to the weights of the samples generated by each
 
double min_distr_wt
 fraction of the total particles that will always be evenly distributed between the samplers;
 
double adaptive_resampling_thresh
 maximum ratio between the number of effective particles and the total particles for resampling to be performed; setting it to <=0 or >1 disables adaptive resampling
 
vectord pix_sigma
 
double measurement_sigma
 
int show_particles
 
bool enable_learning
 
bool jacobian_as_sigma
 
bool debug_mode
 decides whether logging data will be printed for debugging purposes;
 

Member Data Documentation

MeanType PFParams::mean_type

method used for computing the mean of the SSM states corresponding to the particles.

\ 0: No mean computed - just use the state of the particle with the highest weight 1: let the SSM compute the mean of the samples 2: mean of the corners of the bounding boxes corresponding to the particles


The documentation for this struct was generated from the following file: